Half a point. That is what it cost me, the final table put me fourth instead of second, and I have spent five weeks deciding whether to write it up.
Lost is the wrong word if you are strict. The scoresheet says 1/2. But I had a rook endgame that is printed in every endgame primer ever published, a position I have taught to eleven year olds on a magnetic board in a hot classroom in Surulere, and I did not win it. So I am calling it a loss, and I want you to see the exact move.
Skip the first fifty moves. Exchange Slav, I got the better minor piece, he defended sensibly, and by move 54 there was nothing left but this.
Me (White): king g8, rook f1, pawn g7. Him (Black): king e8, rook h2.
My move. Set it up if you have a board, because six moves of geometry follow and prose is a poor board.
My king is standing on the promotion square, in front of my own pawn, which is precisely why nothing has promoted. My rook on f1 owns the f-file, so his king cannot come closer than e8. His rook sits behind me on h2, waiting to check the moment my king steps out.
This is the Lucena position. Not a cousin of it, not a relative. It is the thing itself with the letters shifted along: king in front of its own pawn on the seventh, defending king cut off by a file, defending rook ready to check from behind. The name is a historical accident, incidentally. The earliest surviving discussion is in Alessandro Salvio's Il Puttino of 1634, where Salvio credits Scipione Genovino, and Lucena got his name on it centuries later.
Time control was 90 minutes for 40 moves and then 30 minutes to finish, no increment. I had two minutes forty one seconds.
55. Re1+ Kd7
56. Re4 Rh1
57. Kf7 Rf1+
58. Kg6 Rg1+
59. Kf6 Rf1+
60. Kg5 Rg1+
61. Kf5?? Rxg7
Draw agreed on move 66. I will take the moves one at a time, because each one is doing a specific job and the last one is doing none.
55.Re1+ pushes his king one file further away. This is not a spite check. With his king on e8 my king cannot step to f7, since the two kings would be touching. After 55...Kd7 the square is legal.
56.Re4 is the move that wins the game, and it wins it five moves before anything visible happens. The rook belongs on the fourth rank because my king is going to finish on g5, and a rook dropped onto g4 in front of that king is defended by it. Everything after this is walking.
57.Kf7 comes out from in front of the pawn. Now the checks start and I zigzag: off the file he checked on, but never losing touch with g7. Kg6, Kf6, Kg5. A 1400 player usually asks at this point why I do not just promote. Because the checks would not stop, and a queen on g8 with my king exposed on an open file is a spite queen, not a win.
Then move 61. My king is on g5 and he has just checked along the g-file from g1. Every legal king move loses the pawn: the second my king leaves the g-file, the whole file between his rook and g7 is empty and he takes it for nothing. There is exactly one move.
61.Rg4 is the move. My rook steps in front of my own king, blocking the check, and it stands there protected by the king on g5. He cannot capture: 61...Rxg4+ 62.Kxg4 and the pawn promotes next move with his king two files too far away to help. He cannot check again: the g-file is plugged, and every rank his rook can reach is covered. So 62.g8=Q and the game is over. That is the bridge, and that is why the rook went to the fourth rank on move 56 and not the third, the fifth, or somewhere useful looking.
| Move | My clock afterwards |
|---|---|
| 55 | 2:22 |
| 56 | 2:09 |
| 57 | 1:58 |
| 58 | 1:44 |
| 59 | 1:31 |
| 60 | 1:12 |
| 61 | 0:25 |
Forty seven seconds on move 61. I want to be exact about what I spent them on, because "time trouble" is the story players tell instead of the more embarrassing account.
I was choosing between Kf5 and Kh5.
I was not looking at rook moves at all. My rook had been correct since move 56, it had no further duties as far as I was concerned, and somewhere around move 58 it quietly left the position I was holding in my head. What I had left was a king, a pawn, his rook, and a rhythm. He checks, I step. King, check, king, check. On move 61 the rhythm says king. The position says rook. I obeyed the rhythm, and I did it slowly and with great care.
That failure has a name in every other trade. Dan's write up of forty three minutes of 502s (@grep/postmortem-forty-three-minutes-of-502s) has the same shape: somebody was running a procedure and had stopped reading the screen. The clock did not choose Kf5. The clock stopped me asking which piece moves next, in a position where the answer had just changed.
For the people who ask about the fifty move rule in these endings: it was nowhere near. Article 9.3 lets you claim after fifty moves by each player with no capture and no pawn move, and the count had reset on move 52 when the last pair of pawns came off. The win at move 61 was two moves long.
Here is the part that stings, and it is the part I actually think about.
That endgame was technically lost for him from around move 46, and not because I did anything clever. When the defending king is cut off from the pawn's file, it wants to stand on the short side of the pawn, and the rook wants the long side. With a g-pawn the short side is one file wide, the h-file, and the long side is everything from a to f. The rook needs room for sideways checks to keep coming, four files from the pawn at least, and the king needs to be somewhere other than standing in front of its own rook. That is the whole reason it goes short.
His king spent moves 44 to 54 on d7 and e8. His rook spent them on h2. Both of them on the wrong side of the pawn, exactly reversed. He defended back to front for eleven moves and I handed it back in one.
I have since run the position through a tablebase, which is a slightly humiliating thing to do to yourself. Four pieces on the board, so there is nothing left to have an opinion about. Won before 61.Kf5, drawn after it.
Not "calculate more". That advice sounds like effort and produces nothing.
When I teach the bridge now I make the student say out loud which piece is moving, before their hand leaves the table. King. King. King. Rook. They think the rule is childish until the fourth one, and then somebody's hand goes for the king anyway while their mouth says rook, and they hear it happen. Nobody forgets that.
I showed the position to Chidera on the Saturday. He is 12, rated somewhere around 1180, lying on the floor with a bag of crackers. He found 61.Rg4 in about four seconds and then asked why I had not played it. I said something about the clock. He said I had a whole minute. Correct.